Introduction

Lyme disease (LD), also known as Lyme borreliosis, is an increasingly common vector-borne disease reported in temperate climate zones in North America (NA) and parts of Europe.[1–3] Most human LD infections are caused by three species of bacteria: Borrelia burgdorferi, B. garinii, and B. afzelii [1] Since 2015, LD has been the most common reportable vector-borne disease in NA and Europe.[3,4] Endemic areas in Europe (e.g. Slovenia) and the United States (US) (e.g. Maine) have reported incidence rates of 130 per 100,000 populations in 2010, and 86.4 per 100,000 populations in 2016, respectively. [3,4] Furthermore, current reported rates of LD may be conservative given underreporting estimates of eight to tenfold in the United States. [1] The World Health Organization has made LD a priority disease,[5] as experts predict escalating climate change to play a significant role in the proliferation of this disease due to the expansion of habitable environments for ticks.[6] In Canada, the controversies surrounding the clinical management of LD prompted the federal government to commit to addressing the challenges of recognition, timely diagnosis and treatment of LD, mandated by the unprecedented Federal Framework on Lyme Disease Act.[7]

While mostly treatable, misdiagnosed or untreated LD can result in debilitating long-term sequelae, inappropriate long-term antibiotic therapy and excessive healthcare use.[8] There is currently no human LD vaccine available.[9] The objective of this review was to systematically gather and characterize the body of literature on the economic burden of LD and the cost-effectiveness of LD intervention strategies in order to identify possible knowledge gaps affecting health policy decision-making for LD.

Methods

This scoping review followed the five-step framework by Arksey and O’Malley with guidance from the Joanna Briggs Institute.[10,11] PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines were followed.[12]

Study selection

All search results were aggregated and de-duplicated using Mendeley Reference Management Software. Abstract and title, and full-text screening were completed independently by two reviewers (SM and SDS). Prior to screening, both reviewers conducted a calibration with a set of 100 results. Conflicts at any stage of screening were discussed and resolved through consensus. Disagreements were resolved by a third reviewer (BS). Study selection process and exclusion reasons are shown in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Literature search and study selection.

https://doi.org/10.1371/journal.pone.0210280.g001

Results

Descriptive analysis of economic evaluations

Fig 2 presents an overview of studies categorized by economic evaluation type, geographic region of origin, publication year and impact based on number of Google Scholar citations. The majority of included economic evaluations (n = 11, 52%) were published prior to 2003, with eight CEA[17–24] and three cost analyses.[27–29] From 2003 and onwards, there were 10 published economic evaluations: two CEA, [25,26] and eight cost analyses. [30–37] All LD intervention CEAs were from NA countries, while cost analyses were published more frequently after 2003, and from European countries. The cost-effectiveness of antibiotic treatment strategies,[17] and a diagnostic test cost analysis,[34] were considered most impactful studies based on their number of Google Scholar citations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Bubble chart displaying studies included in review by study year, geographic region, and type of economic evaluation.

Balloon size depicts the number of citations through Google Scholar.

https://doi.org/10.1371/journal.pone.0210280.g002

Data source usage

Data sources used by the 21 economic evaluations, stratified by CEA and cost analysis, were summarized in Fig 3. Literature use was most common (n = 10, 48%), followed by insurance claims information (n = 8, 38%). CEAs mostly used existing literature, reports, and consulting experts, while cost analyses mostly used insurance claims or health provider data as well as questionnaire data to complete their economic evaluations. There were no economic evaluations that used health administrative data from a single payer health system.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Data sources used by LD cost-effectiveness analyses and cost analyses.

* New Jersey Blue Cross—Blue Shield; Diversified Pharmaceutical Services; Delmarva Health Plan; Swedish Social Insurance; German DAK; IMS Health LifeLink Health Plan Claims; ** CDC LD Incidence Reports, Epidemiologic Reports, Lyme Disease Vaccine Study Group; *** Scottish Health Service, Departments of Economy and Information Technology.

https://doi.org/10.1371/journal.pone.0210280.g003

Economic evaluations assessing cost-effectiveness

Study characteristics.

Study design characteristics are summarized in Table 1. Six CEA studies used a healthcare payer perspective, [17–20,24,25] and four studies used a societal perspective,[21–23,26] of which three compared the cost-effectiveness of a vaccination program.[21–23] One study also used a hospital perspective in the sensitivity analysis.[26]

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Economic evaluation study characteristics.

https://doi.org/10.1371/journal.pone.0210280.t001

The time horizon ranged between one-year and lifetime, with seven studies using a time horizon less than 10 years, [17,18,22–26] and only two using a lifetime horizon.[19,20] Ninety percent of CEAs completed a deterministic sensitivity analysis.[17–25] Probabilistic sensitivity analyses were not presented. Discounting was used in six studies, [19–24] where rates varied between 3% and 5% for base-case analyses, and between 0% (i.e. no discounting) and 10% in sensitivity analyses.

Model type and interventions assessed.

Seven CEAs reported an expected cost-effectiveness outcome using decision tree analysis [17,18,20–22,25,26];three CEAs used a Markov state-transition model.[19,23,24]. Four studies assessed the cost-effectiveness of vaccination.[21–24] Five CEAs assessed antibiotic cost-effectiveness: three studies compared treatment algorithms for early and late disseminated sequelae, [17,19,20] one compared treatment algorithms for early localized LD, [25] and one assessed cost-effectiveness of intravenous ceftriaxone in patients who lacked classical clinical manifestations.[18]

Outcomes.

Cost, health, and cost-effectiveness outcomes reported are summarized in Table 2. Types of direct costs (healthcare costs) incorporated into models included: laboratory diagnostics, physician services, hospital care, medication and associated-adverse events (e.g. antibiotic treatment), sequelae, vaccination (e.g. administration, time, travel), and cost per LD case. Indirect costs included productivity loss.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Primary study characteristics and conclusions of cost-effectiveness analyses.

https://doi.org/10.1371/journal.pone.0210280.t002

Health outcomes chosen for CEAs included: the number of major and minor complications (sequelae), number of therapy-related adverse events, number of LD cases averted, life expectancy, QALYs, and mortality. One study used test sensitivity and specificity outcomes. [26]

Incremental cost-effectiveness ratios (ICERs) reported were: cost per additional major complications prevented, cost per late LD case prevented, cost per QALY, and cost per LD case averted. One study did not conclude a cost-effectiveness ratio outcome.[26]

Study findings

Conclusions from all CEA are summarized in Table 2. All ICERs were inflated and standardized to 2017 USD per QALY. The ICER for vaccination programs are summarized in Table 3 and ranged between 7,024 USD (probability of LD infection of 0.5%, [21]) and 2.36M USD (probability of LD infection of 0.0067%, [24]) per LD case averted. Studies reporting the ICER in USD per QALY reported results between 93,619 (probability of infection of 1%, [23]) and 5.17M (probability of infection of 0.0046%, [22]) USD per QALY. The ICER varied depending on the probability of LD infection, probability of diagnosing early LD and vaccination costs. These three CEAs, all from societal perspectives, concluded that vaccination was likely economically favorable for endemic LD areas and not cost-effective for nation-wide administration. All four studies used a time horizon of 10 to 11 years, performed sensitivity analyses, and discounted at 3%.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Summary of standardized ICER for vaccination programs in the United States.

https://doi.org/10.1371/journal.pone.0210280.t003

In 1992, Majid et al. concluded that empirical antibiotic treatment of patients with tick bites was cost-effective when the probability of infection was 0.036 or higher.[17] Subsequent studies by Lightfoot et al., Nichols et al., and Lantos et al., also reported that an empirical antibiotic approach is cost-effective and preferred for patients with a positive Lyme antibody titer, if the pretest probability for LD is high, and for patients in regions endemic for LD.[18,20,25] The study by Eckman et al. assessed the cost-effectiveness of oral antibiotic treatment using 100 mg of doxycycline compared to an intravenous administration of 2 g of ceftriaxone. This study concluded that oral doxycycline was dominant (cost savings and provided an additional 0.1 QALY) in both early LD and Lyme arthritis patients.[19]

Economic evaluations assessing LD-associated costs

Study characteristics.

Seven studies assessed the economic burden of LD using total healthcare costs,[28–31,35–37] three studies included diagnostic testing costs only,[27,32,34] and one study included Lyme cardiac treatment costs only.[33] Five studies used a healthcare payer perspective,[27,32–35] six studies used a societal perspective, [28–31,36,37] and two studies used a third-party payer perspective.[28,36] Approximately 91% (10 of 11) of cost analyses used a time horizon between 0 and 5 years.[27–36] Only four studies completed any form of sensitivity analysis, [29,33,35,37] and two studies used discounting with rates between 3 and 4%.[28,37]

Outcomes.

Outcomes reported are summarized in Table 4. Cost analyses focusing on overall healthcare costs included direct medical costs: outpatient visits and related healthcare utilization, hospitalizations, emergency room visits, home health care, prescription medication (antibiotic treatment), cost of subsequent manifestations (major or minor sequelae), consultations, laboratory costs, and treatment side effects. Diagnostic cost analyses only included serologic test costs and laboratory costs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Primary study characteristics and conclusions of cost analyses.

https://doi.org/10.1371/journal.pone.0210280.t004

Indirect costs that were incorporated for societal perspectives included: out-of-pocket drug costs, caregiving, travel, work loss, restricted-activity days at home, and loss of healthy time from sequelae. The study by Joss et al. incorporated the cost in the management of patients (e.g. consultations and screening test costs) found not to have evidence of the disease to evaluate non-confirmed LD patient burden from a societal perspective. [29] Two studies used a human capital approach,[30,36] one study used friction cost methods, [37] and three studies used secondary data to estimate indirect costs. [28,29,31]

Study findings.

Three diagnostic cost analyses were included in this review (Table 4). Reported costs are standardized to 2017 USD currency using its respective inflation. The first US study by Strickland et al. concluded that 30,000 tests for LD were performed annually on Maryland residents, totalling direct medical costs of over 3.23M USD.[27] More recently, Hinckley et al. concluded that 3.4 million LD tests were conducted by the seven laboratories involved in their study (from four endemic states: Connecticut, Maryland, Minnesota and New York), at an estimated national cost of 566M USD.[34] Both studies concluded that diagnosis costs are a concern and should be included in the public health burden of LD. In Europe, a study by Muller et al. concluded that the overall expected cost of diagnostic testing and treatment was estimated at 67.93M USD in Germany, and suggested a high amount of potentially inappropriate healthcare services utilized for patients with a suspected or confirmed diagnosis of Lyme borreliosis.[32]

Overall, there were six cost analyses that assessed the economic burden of LD from a societal perspective: four from Europe, and two from the US (Table 4). Joss et al. reported an annual national economic burden of 735,550 USD (0.14 USD per capita, n = 5.40M) for Scotland,[29] while Henningsson et al. reported a national economic burden of 712,808 USD over 5 years in Sweden (0.07 USD per capita, n = 9.96M) for neuroborreliosis-related healthcare.[31] Furthermore, Lohr et al. reported an annual national economic impact of over 40.88M USD in Germany (0.51 USD per capita, n = 80.59M),[36] and van den Wijngaard et al. reported an annual national cost of 23.12M USD for LB in the Netherlands (1.36 USD per capita, n = 17.08M).[37]

In the US, cost analyses were completed sporadically from 1998 to 2015. A cost-of-illness study by Maes et al. reported an expected national expenditure of 3.93 billion USD over five years (2.41 USD per capita per year, n = 326.63M).[28] A healthcare utilization study by Zhang et al. reported direct medical costs of 4,273 USD and indirect costs of 7,484 USD per LD patient, totalling an estimated nationwide economic impact of 292M USD (0.89 USD per capita, n = 326.63M).[30] Similarly, a recent study by Adrion et al. reported an additional 3,084 USD of healthcare costs per LD patients over a 12-month period. They also determined that persistent LD sequelae (post-treatment LD syndrome, PTLDS) are associated with an increase of 3,946 USD healthcare costs compared to patients without PTLDS.[35]

Discussion

We summarized a total of 21 economic evaluations (10 CEA and 11 cost analyses) related to LD. The majority of CEA studies were conducted prior to 2003, which was related to the previously available LD human vaccine. [9] Since the vaccine was withdrawn, there has been no novel intervention strategies for LD and subsequently minimal interest in CEA studies after 2003. Although this vaccine was withdrawn for reasons other than cost-effectiveness, [9] all four LD vaccination CEAs concluded that universal vaccination in the US was likely not cost-effective.

We included seven cost analyses focused on overall healthcare costs, three studies focused on diagnostic testing and one cost analysis focused on Lyme cardiac treatment. A common theme of the diagnostic cost analyses was the burden of inappropriate and over-usage of LD diagnostic testing in the US and Germany. While diagnostic economic evaluations specifically looking at costs are appreciated, it would be difficult for decision-makers to use this evidence in the absence of overall healthcare burden. Of the seven cost analyses assessing burden through total healthcare costs, three European studies concluded that further research and priority should be placed on preventive interventions for LD. Based on the most recent study by Zhang et al, the inflated annual economic impact for LD in the US was 292M USD. While this does not come close to the burden of influenza, cancer or chronic conditions (e.g. diabetes, obesity), it falls in the same magnitude of other high-profile vector-borne diseases in the US such as West Nile virus (778M USD over 13 years),[38] and Zika virus (500M USD annual assuming a 0.3% attack rate across six prominent states).[39] Overall, the economic burden of LD could be considered significant to the US and other developed countries to justify further research efforts in LD control and management.

There are limitations to this review, as resource constraints limited our literature search to articles written in English, introducing possible language bias. As a result, there may be an underrepresentation of European studies, which should not be interpreted as a lack of interest or lower LD incidence rates in this region. We did not attempt to identify costs associated with LD avoidance (i.e. non-health related prevention) since we were interested in the economic burden of LD on the healthcare system and society. Per capita costs were presented by dividing nation-wide burden by the entire population. However, it should be acknowledged that not everyone from a specific country are susceptible to LD. As a result, our review may be underestimating the actual per capita costs in high incidence areas, and we advise against using these per capita estimates to describe LD burden. Lastly, since the goal of this scoping review was to characterize the literature, risk of bias assessment and quality appraisal were not completed. We propose quality appraisal of the literature be explored in a future systematic review.

To our knowledge, this was the first study that systematically identified and characterized the economic evaluation literature for LD. In 1999 and 2002, reviews by Rouf et al. and Tella et al. identified costs, and cost-effectiveness studies in rheumatology, respectively.[40,41] However, both reviews identified limited LD studies and were not able to provide a comprehensive description of the burden of LD. The search strategy was comprehensively designed and adapted to four electronic databases to search NA and European literature. Given the amount of HTA and health economics organizations that release reports on vector-borne diseases, our search in the grey literature added to the comprehensiveness of this review. The timing of this review should be useful for health services and LD researchers alike aiming to understand the implications of this emerging infectious disease where it is estimated that 300,000 cases of LD are diagnosed annually in the US,[42] with limited development of novel interventions. [43]

A recent scoping review from Canada by Greig et al. identified all LD literature (e.g. risk factors, surveillance, diagnostics) related to public health. In this review, they identified 32 hits related to economic burden of LD or cost-benefit of interventions, but do not specifically report on the results, trends or conclusions of the studies. [44] In comparison, we included fewer studies since we excluded abstracts, editorials, secondary reviews, and economic evaluations not directly related to Lyme disease. Our review comes to a similar conclusion in that economic burden studies for LD are limited.

Our review was able to highlight specific research gaps in the LD literature. Of the 11 cost analyses, six studies reported societal costs (i.e. productivity loss, indirect costs, non-medical costs) between 23 and 64% of total economic costs. However, many of these indirect costs were roughly estimated using friction cost or human capital approach methods. It is evident that while healthcare costs are significant for LD in various countries, the societal costs are equally as impactful for this disease and should be further studied. Our review also summarized the range of economic impact across various countries known to have increasing rates of LD, and countries that have not estimated the economic impact of this vector-borne disease while facing increasing LD cases (e.g. Canada)[45] or historically have high LD incidence rates (e.g. Slovenia, Czech Republic).[46] Future efforts in identifying specific LD stages, indirect costs, or healthcare utilization that create the highest economic burden can be useful to support public health agenda in countries with this vector-borne disease.

There was a high degree of heterogeneity in economic evaluation methods, data sources and outcomes reported. The cost/QALY gained outcome is typically used to express the cost-effectiveness to health policy decision-makers, since it can be compared to commonly-used thresholds (e.g., $50,000/QALY,[47] and 20,000 Sterling Pounds/QALY in the United Kingdom [48]). However, many studies reported cost-effectiveness in other units, limiting appropriate comparisons. We also noticed an array of LD health states, and health state utility values (HSUV) used. HSUVs for LD health states were mostly derived from expert clinical opinion, which could in turn be underestimating the QALYs and the cost-effectiveness of interventions. Furthermore, an individualized approach (e.g. individual-level microsimulation) may be more accurate in predicting cost-effectiveness of LD interventions, since unique baseline characteristics of patients (e.g. comorbidities and demographics) can affect disease progression and subsequently predicted lifetime outcomes.

Only cost analyses from the US provided sequelae-attributable costs and case-attributable costs per patient.[35] As big data and computing power evolve in health care, future studies can further investigate attributable healthcare costs using health administrative data to determine population-specific burden. Future health services research should thus consider the local context in generating evidence to support health decision makers given the regional differences in LD incidence, detection, symptoms, sequelae and healthcare systems.

Conclusions

This scoping review identified 21 economic evaluations for Lyme disease from North America and Europe. Similar to other vector-borne diseases, the burden of Lyme disease suggests an economic argument for further research. A greater understanding of the indirect costs of Lyme disease and cost-effectiveness of interventions in countries where the incidence rates of the disease are increasing, is warranted for guiding Lyme disease evidence-informed health policy decision making.

Supporting information

Acknowledgments

The authors wish to acknowledge librarian Vincci Lui for her guidance on the best practices of literature search strategy development.